Граматика / English Syntax

Now draw the tree for (i) above and also for (ii)–(iv) below. Use arrows to indicate movement. Remember that carefully and accidentally are VP-Adjuncts, while probably is an S-Adjunct.

(ii)Edward will have carefully wrapped the present.

(iii)He may have broken the vase accidentally.

(iv)Chuck will probably not have seen it.

*10. When we discussed Wh-movement in the text, we looked only at Whphrases that are arguments. Consider the sentences below where (ii) can be said to be derived from (i). Draw the trees for both sentences.

(i)You can eat pancakes where?

(ii)Where can you eat pancakes — ?

*11. Describe the movement(s) in the following sentence; then draw the tree:

(i)What has he eaten?

Further Reading

Movement is a notion found only in transformational grammar (TG), a theory of language associated with the linguist Noam Chomsky. In this theory (with the exception of the latest version) elements move from a

D(eep)-Structure level of representation to a S(urface)-Structure level.

In TG there is a special position for moved Wh-elements, called COMP, or simply ‘C’, which normally hosts complementisers. As I already mentioned at the end of the previous chapter, this position is the Head of a maximal projection called CP. For further discussion of movement processes from a slightly di erent perspective than the one taken here, see Radford (1988), Haegeman (1994) and Ouhalla (1999).

This page intentionally left blank

Part III

Argumentation

This page intentionally left blank

You observe that words like the and my, just like warm and tall, can occur before nouns, and have some sort of modifying function. You now surmise that, by analogy with warm and tall, words which you know to be adjectives, the and my are also adjectives. However, you realise that it is not wise to base your conclusions on only a few examples, and at this stage you want to refine your hypothesis and look for further data that bear on the question of the categorial status of the, my, etc. In fact, what you want to do is progressively falsify your hypotheses by finding counterexamples to them, and in this way continually adjust your initial suppositions. Returning to the problem at hand, let us consider some more data:

(5)the warm sunshine

(6)my tall palmtrees

(7)*my the sunshine

(8)*the my palmtrees

These examples are problematic for the initial hypothesis that warm, tall, the and my are all elements from the same word class. If they were, why is it that warm and tall can be preceded by the or my, i.e. an item from our mystery category (cf. (5) and (6)), but that two items from this category cannot cooccur (cf. (7) and (8))? There must be something that distinguishes words like warm and tall from words like the and my. Consider next the following phrases:

(9)beautiful, warm, southern sunshine

(10)*the my some island

Clearly elements that are indisputably adjectives can be stacked (see Section 7.2), but the same cannot be said for our mystery elements: like (7) and (8),

(10)shows that only one of these can be selected. It soon becomes obvious that there are many more examples where there are di erences between warm and tall on the one hand, and the and my on the other:

(11)very warm sunshine

(12)extremely tall palmtrees

(13)*very the sunshine

(14)*extremely my palmtrees

These examples show that words like warm and tall can be preceded by an intensifying adverb like very or extremely (see Chapter 3), while the and my cannot. In addition, compare (15) and (16):

(15)Sunshine is warm.

(16)*Sunshine is the.

It is now clear that on the basis of even a handful of data the initial hypothesis that words like the and my are adjectives must be abandoned.

The next stage is to come up with a new hypothesis. Given the facts above, it would be reasonable to surmise that words like the, a, my, etc. belong to a di erent word class than the class of adjectives. As we saw above, this new class consists of elements that we have been calling determiners. We have also seen that at any one time we can select only one determiner. At least, that’s what the data we’ve looked at so far suggest.

Having been on the island for quite a while now, you will have had time to think of plenty more relevant examples, and you come up with (17):

(17)All my many good ideas to get o this island have failed.

What’s interesting about this example is that apparently in the italicised string we have three determiners, namely all, my and many. So now we’re faced with a situation in which in some cases it’s not possible to have more than one determiner (cf. (7), (8) and (10)), but in other cases it is (cf. (17)).

We’re now dealing with quite complex structures, and if you think about it, (17) is not only problematic for lone mariners, but for professional grammarians as well. The reason is that the NP we have in (17) cannot easily be accommodated in the X0-theory that we have adopted in this book: after all, in (17) we have three determiners, but in NPs there is only one Specifier position, which is where determiners are located (cf. Section 7.1). Recall that adjectives that occur in NPs are not problematic in this way. Functionally they are Adjuncts which can recursively be stacked by creating new bar-level nodes (see Section 7.2). The structure of the Noun Phrase in English is a di cult area of grammar, and the subject of much current research, so we won’t attempt to provide a definitive analysis of English NPs.

For our purposes what is most important is that you now have an idea of how argumentation proceeds: it can be seen as an ongoing process of hypothesis refinement by taking into account more and more linguistic data. Where do these data come from? We can either construct them from our knowledge of the language (these are called introspective data), or we can collect them from what we hear around us in conversations, on radio or television, or from what we read in newspapers, books, etc. (these are called attested data). There is some debate about the question of which kind of data are the most valuable, but the common-sense view is to use any data that are relevant to our concerns.

It’s important to stress that our reasoning should be systematic and informed. This is not to say that a certain amount of intelligent guesswork is not part of the argumentative process: especially in the early stages, after first encountering a problem concerning syntactic analysis, we may well find ourselves guessing how we can resolve it. We may even come up with a

number of alternative analyses. What is then needed is a procedure for finding support for the conjectured analysis, or for making choices between alternatives.

We now turn to the question what sort of arguments we can use to evaluate a proposed analysis of a particular construction, and how we can choose between rival analyses.

10.2Economy of Description: Linguistically Significant Generalisations and Occam’s Razor

In this section I will discuss two ways in which economy of description should play an important role in analysing a syntactic construction, or in choosing between two or more competing analyses of some phenomenon. It will be intuitively obvious to you that we should rate highest the simplest analysis that successfully accounts for the data, everything else being equal. Simplicity of description can be achieved in two ways: on the one hand by making Linguistically Significant Generalisations, and on the other hand by reducing our terminological repertoire.

10.2.1Linguistically Significant Generalisations

We start with the notion ‘generalisation’. If we’re engaged in describing a complex system, any complex system, our task is made easier if we can organise the data at our disposal in a systematic way. A systematic description is called a taxonomy. Perhaps the most famous taxonomist was the Swedish naturalist Linnaeus (1707–78), who set up taxonomies of animals and plants. It will be obvious to you that Linnaeus did not achieve fame by providing a random catalogue of animals and plants and a list of their characteristics. He organised the data into patterns in such a way that generalisations could be made. It would take us too far afield to discuss Linnaeus’s taxonomy in any sort of detail, but we can get some idea of his system by briefly considering one type of animal, namely cats. We all know that domestic cats, tigers and lions belong to the animal family of felines. If, in setting up a taxonomy, we were to classify each of these animals in a class of their own we would miss the generalisation that they all belong to the family of cats by virtue of certain shared physical characteristics, e.g. the shape of their ears, the fact that they have whiskers, etc.

The considerations above regarding animal taxonomies also apply when we attempt to describe the grammar of English. Here too it is important to realise that we don’t want a taxonomy in which each element of the language is classified individually. What we want instead is a descriptive system in which we can draw analogies between elements and categories on the basis of their shape and distributional properties. In other words, we

want a tightly organised description in which maximum use is made of generalisations. However, we are not interested in any generalisation, only in what have been called Linguistically Significant Generalisations (LSGs). These are generalisations that are significant to the extent that they express regular patternings observed in a particular language or across several languages. We now turn to some examples of LSGs.

You may not have realised it, but in previous chapters we have already made extensive use of generalisations. For example, when we looked at word classes in Chapter 3, we grouped words together on principled grounds, namely on the basis of shared syntactic, and in some cases, morphological characteristics. So even at that stage we were very much engaged in an argumentative process. Argumentation also played an important role when we discussed the internal structure of phrases in Chapter 7. There we introduced X-bar syntax as a system that achieves cross-categorial generalisations, and we saw that the structure of all phrases in English can be described as in (18):

X0 (Adjunct)

Head/X (Complement)

Each phrase XP (where X ¼ N, V, A, P) contains a Specifier whose sister constituent is a bar-level category. Adjuncts (if present) are sisters of bar-level categories, while Complements (if present) are sisters of lexical categories. As a result of the insights o ered by the cross-categorial generalisations of X-bar theory, our description of English syntax is considerably neater and tidier than a system which proposes flat structures for phrases would ever be. The reason is that the need to describe the skeletal structure of each phrase individually is obviated.

Cross-categorial generalisations can also range over a smaller set of categories. Consider the sentences below and their variants:

(19)Kate came to see me. — It was Kate who came to see me.

(20)I met her in Philadelphia. — It was in Philadelphia that I met her.

(21)I made her work. — *It was work that I made her.

(22)I made her happy. — *It was happy that I made her.

The second sentences in each case are called cleft sentences. They take the following form:

It þ form of be þ focus (italicised in the sentences above) þ who/that . . .